RD Hunyuan Video lora

Lora for Hunyuan Video trained on RD in 2D and 3D.

Activation keyword is RDLORA, You can just do pony Rainbow Dash doing whatever. Your going to want at least 3 seconds worth of frames imo, higher fps = more movement but also more gen time since you’ll need more frames obviously. 512 x 512 seems to work pretty well. Use fastvideo lora and you can get decent results with 6 steps. Might do one for all of the mane six later if I find the patience again.

If you want to use it with other loras you might have to upweight the pony Rainbow Dash part.

Note: E6 is the latest

I included the workflow, for it to work you will need these nodes:

https://github.com/city96/ComfyUI-GGUF 
https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite
https://github.com/Lightricks/ComfyUI-LTXVideo (I know its LTX its to use STG in it.)

And I use this clip instead of the regular one:
https://huggingface.co/zer0int/CLIP-SAE-ViT-L-14/blob/main/ViT-L-14-GmP-SAE-FULL-model.safetensors 

Here is everything else:

https://huggingface.co/Kijai/HunyuanVideo_comfy/blob/main/hyvideo_FastVideo_LoRA-fp8.safetensors
https://huggingface.co/city96/HunyuanVideo-gguf/blob/main/hunyuan-video-t2v-720p...
https://huggingface.co/Comfy-Org/HunyuanVideo_repackaged/tree/main/split_files/vae